Availability of Rapid Human Immunodeficiency Virus Testing in Academic Emergency Departments

Authors

  • Peter D. Ehrenkranz MD, MPH,

    1. From the Robert Wood Johnson Clinical Scholars Program (PDE, JPM), the Departments of Medicine and Biostatistics and Epidemiology (PDE, JPM, WCH), and the Centers for Education and Research on Therapeutics (PDE, JPM), University of Pennsylvania, Philadelphia, PA; the Leonard Davis Institute for Health Economics (PDE, JPM, WCH) and the VA Medical Center (JPM, WCH), Philadelphia, PA; the Department of Emergency Medicine, Massachusetts General Hospital (CJA, CAC), Boston, MA; and the Department of Emergency Medicine (RR), Johns Hopkins Medical Center, Baltimore, MD.
    Search for more papers by this author
  • Christina J. Ahn BA,

    1. From the Robert Wood Johnson Clinical Scholars Program (PDE, JPM), the Departments of Medicine and Biostatistics and Epidemiology (PDE, JPM, WCH), and the Centers for Education and Research on Therapeutics (PDE, JPM), University of Pennsylvania, Philadelphia, PA; the Leonard Davis Institute for Health Economics (PDE, JPM, WCH) and the VA Medical Center (JPM, WCH), Philadelphia, PA; the Department of Emergency Medicine, Massachusetts General Hospital (CJA, CAC), Boston, MA; and the Department of Emergency Medicine (RR), Johns Hopkins Medical Center, Baltimore, MD.
    Search for more papers by this author
  • Joshua P. Metlay MD, PhD,

    1. From the Robert Wood Johnson Clinical Scholars Program (PDE, JPM), the Departments of Medicine and Biostatistics and Epidemiology (PDE, JPM, WCH), and the Centers for Education and Research on Therapeutics (PDE, JPM), University of Pennsylvania, Philadelphia, PA; the Leonard Davis Institute for Health Economics (PDE, JPM, WCH) and the VA Medical Center (JPM, WCH), Philadelphia, PA; the Department of Emergency Medicine, Massachusetts General Hospital (CJA, CAC), Boston, MA; and the Department of Emergency Medicine (RR), Johns Hopkins Medical Center, Baltimore, MD.
    Search for more papers by this author
  • Carlos A. Camargo Jr. MD, DrPH,

    1. From the Robert Wood Johnson Clinical Scholars Program (PDE, JPM), the Departments of Medicine and Biostatistics and Epidemiology (PDE, JPM, WCH), and the Centers for Education and Research on Therapeutics (PDE, JPM), University of Pennsylvania, Philadelphia, PA; the Leonard Davis Institute for Health Economics (PDE, JPM, WCH) and the VA Medical Center (JPM, WCH), Philadelphia, PA; the Department of Emergency Medicine, Massachusetts General Hospital (CJA, CAC), Boston, MA; and the Department of Emergency Medicine (RR), Johns Hopkins Medical Center, Baltimore, MD.
    Search for more papers by this author
  • William C. Holmes MD, MSCE,

    1. From the Robert Wood Johnson Clinical Scholars Program (PDE, JPM), the Departments of Medicine and Biostatistics and Epidemiology (PDE, JPM, WCH), and the Centers for Education and Research on Therapeutics (PDE, JPM), University of Pennsylvania, Philadelphia, PA; the Leonard Davis Institute for Health Economics (PDE, JPM, WCH) and the VA Medical Center (JPM, WCH), Philadelphia, PA; the Department of Emergency Medicine, Massachusetts General Hospital (CJA, CAC), Boston, MA; and the Department of Emergency Medicine (RR), Johns Hopkins Medical Center, Baltimore, MD.
    Search for more papers by this author
  • Richard Rothman MD, PhD

    1. From the Robert Wood Johnson Clinical Scholars Program (PDE, JPM), the Departments of Medicine and Biostatistics and Epidemiology (PDE, JPM, WCH), and the Centers for Education and Research on Therapeutics (PDE, JPM), University of Pennsylvania, Philadelphia, PA; the Leonard Davis Institute for Health Economics (PDE, JPM, WCH) and the VA Medical Center (JPM, WCH), Philadelphia, PA; the Department of Emergency Medicine, Massachusetts General Hospital (CJA, CAC), Boston, MA; and the Department of Emergency Medicine (RR), Johns Hopkins Medical Center, Baltimore, MD.
    Search for more papers by this author

  • None of the authors have financial interests, relationships, or affiliations relevant to the subject of the manuscript. The work was funded by the Agency for Healthcare Research and Quality and its support of the Center for Research on Therapeutics at the University of Pennsylvania, as well as the Robert Wood Johnson Foundation and its support of the Clinical Scholars Program.

Address for correspondence and reprints: Joshua P. Metlay, MD, PhD; e-mail: jmetlay@mail.med.upenn.edu.

Abstract

Objectives:  The Centers for Disease Control and Prevention (CDC) recommends routine human immunodeficiency virus (HIV) screening of emergency department (ED) patients aged 13 to 64 years. The study objectives were to determine the accessibility of rapid HIV testing in academic EDs, to identify factors that influence an ED’s adoption of testing, and to describe current HIV testing practices.

Methods:  Online surveys were sent to EDs affiliated with emergency medicine (EM) residency programs (n = 128), excluding federal hospitals and facilities in U.S. territories. Eighty percent (n = 102) responded. Most e-mail recipients (n = 121) were Emergency Medicine Network (EMNet) investigators; remaining contacts were obtained from residency-related Web sites.

Results:  Most academic EDs (n = 58; 57%; 95% confidence interval (CI) = 47% to 66%) offer rapid HIV testing. Among this group, 26 (45%) allow providers to order tests without restrictions. Of the other 32 EDs, 100% have policies allowing for rapid HIV testing following occupational exposures, but less than 10% have guidelines for testing in other clinical situations. Forty-seven percent expect to routinely offer HIV testing in the next 2 to 3 years. Only 59% of the EDs that offered rapid tests in any situation could link an HIV-positive patient to subspecialty care. The facility characteristic most important to availability of rapid HIV testing was the presence of on-site HIV counselors.

Conclusions:  Most academic EDs now offer rapid HIV testing (57%), but few use it in situations other than occupational exposure. Less than half of academic EDs expect to implement CDC guidelines regarding routine screening within the next few years. The authors identified facility characteristics (e.g., counseling, ability to refer) that may influence adoption of rapid HIV testing.

Almost 1 million Americans are infected with human immunodeficiency virus (HIV), yet as many as 250,000 of them do not know their serologic status.1 Increasing the availability and utilization of HIV testing would result in identification of undiagnosed carriers. These individuals would then have the opportunity to commence medical care and would benefit from appropriate use of antiretroviral medications and prophylaxis against opportunistic infections. Such therapies have been shown to lower viral loads, reduce the likelihood that the virus will be transmitted to others, decrease morbidity and extend life expectancy of the infected patient.2,3

In September 2006, the Centers for Disease Control and Prevention (CDC) issued recommendations encouraging providers in all health care settings to perform screening for HIV infection “routinely for all patients aged 13–64 years.”4 The recommendations note that emergency departments (EDs) are an underutilized venue for nontargeted population-based HIV testing. They advise facilitating screening efforts at these sites through the use of rapid HIV tests instead of standard serum-based assays. However, EDs have traditionally been reluctant to participate in preventive health initiatives,5 and the CDC guidelines have led to controversy within the emergency medicine (EM) community.6–10 Although no standard of care or professional consensus has emerged, the CDC has taken an active stance to promote implementation, fostering partnerships between local health departments and EDs, as well as offering ED-based workshops aimed at implementing routine HIV screening.

To provide better direction to policy makers and physicians seeking to promote routine nontargeted HIV screening in EDs, we conducted a national survey on rapid HIV testing practices in academic EDs. For the purpose of this study, we defined and used the term “testing” in the broadest possible context, as performance of an HIV assay for any reason whatsoever, ranging from diagnosis of acute illness to routine screening of asymptomatic persons. “Screening,” on the other hand, is restricted to those instances in which HIV assays are performed on a population level (i.e., offered to all patients in a population regardless of signs, symptoms, or presence of risk factors for HIV).

Our primary objective was to identify the hospital-level and regional factors that influence an academic ED’s use of rapid HIV tests. The secondary objective was to assess HIV testing policies and practice patterns among respondent EDs and to compare these data with data from a similar study conducted in 1996. General attitudes toward routine HIV screening and its expected effect on patient flow through the ED among respondent ED physicians were also assessed.

Methods

Study Design and Population

We conducted a cross-sectional survey of participants in the Emergency Medicine Network (EMNet) affiliated with U.S. academic EDs between December 2006 and March 2007. The study was designed as a follow-up to an earlier survey of HIV testing practices in academic EDs that was conducted by a different group of investigators in 1996, prior to widespread marketing of rapid HIV tests.11 We surveyed an EMNet respondent from each of the 128 adult EDs associated with accredited allopathic EM residency training programs at U.S. academic medical centers. We specifically chose centers with affiliated residency programs because we believed that these would be most likely to adopt new technologies and recommendations first. We excluded federal hospitals and facilities in United States territories (n = 7) because we expected that they would be governed by different cultural and statutory norms. Each of the surveyed programs is located at a hospital in the United States that is accredited by the Residency Review Committee of Emergency Medicine (RRC-EM) and the Accreditation Council of Graduate Medical Education. The survey questions were directed at policies and practices of the primary affiliated hospital. This sample of sites is identical to the one surveyed in 1996, although the specific respondents are not identical.11 The study was approved by the institutional review board at each of the authors’ institutions.

Survey Content and Administration

We identified a primary, secondary, and tertiary emergency physician, researcher, or administrator to contact from each site using the EMNet. The vast majority of prospective participants (n = 121) were physicians who represent the sites within EMNet, that participate in ongoing clinical research studies, and have previously responded to similar questionnaires about ED performance.12,13 EMNet is an academic research network that focuses on public health issues that affect the delivery and quality of emergency care, as well as the primary care needs of medically disadvantaged populations (http://www.emnet-usa.org/). The network is coordinated at the Massachusetts General Hospital in Boston, Massachusetts. Its current membership includes 148 medical centers in 32 states, many of which are academic centers; those that do not have academic residency programs were excluded from this study (n = 20). For the seven sites with academic EM residency programs that are not in EMNet, we searched the Society for Academic Emergency Medicine website to identify the residency program director or department chair.

Each prospective respondent received up to three e-mails with a link to an Internet-based questionnaire that was created and administered by EMNet via EZSurvey (http://www.ezsurvey.com/). If the first contact from a site did not respond after three e-mails, the questionnaire was sent to the preidentified secondary contact within the ED and then, if necessary, to the third. A site identification number submitted on return of the questionnaire was used to track which EDs completed the questionnaire and to enable linking questionnaire data to other, public-use data about participating sites.

The survey instrument was assembled from the published literature: it was our intention to reassess questions asked 10 years earlier by Wilson et al.11 A focus group performed by the CDC of emergency providers offered additional themes on barriers to ED-based HIV testing.14 We decided a priori not to identify formal domains or perform dimension-level reliability testing since the survey was intentionally kept short and asked predominantly concrete questions relating to facility resources or policies.

Our instrument contained 21 items listing facility-level and respondent-specific characteristics potentially associated with HIV testing. A link to the questionnaire can be found at http://www.emnet-usa.org/cgi-bin/ezs.exe?DATABASE=hiv_survey. Respondents indicated level of agreement with each item using dichotomous responses or 5-point Likert scales. Three of the questions asked respondents for their opinion. Two of these inquired about the impact of aspects of rapid HIV testing on the flow of patients through the ED. The third asked about potential plans for routine HIV screening.

In addition to the data on facility characteristics, we used the questionnaire to ask respondents about HIV counseling and testing policies and typical practices at their ED in six specific clinical scenarios: a patient with unknown HIV status presenting with 1) an occupational exposure to HIV; 2) clear signs of an acquired immune deficiency syndrome (AIDS)-defining illness requiring admission; 3) clear signs of an AIDS-defining illness not requiring admission; 4) an uncomplicated suspected sexually transmitted disease (STD); 5) an overdose from intravenous opiates requiring admission; and 6) history of a motor vehicle collision. After presenting each scenario, we asked the respondents to provide information about their ED’s policies or guidelines regarding risk reduction counseling, rapid or standard HIV testing in the ED, or written referral for HIV testing after discharge.

One section of the questionnaire asked the respondent to rank the three factors that she or he felt were most important to the accessibility of rapid HIV testing in their ED from a list of 13 potential response options. For each factor, we report the percentage of respondents who ranked it among their top three items of importance.

The questionnaire was pilot tested by the EMNet Steering Committee and with 20 EM faculty and residents at the Hospital of the University of Pennsylvania to assess feasibility, comprehensibility, and anticipated response rates. Characteristics not included in the questionnaire, such as AIDS prevalence by state, state-specific counseling and testing laws, and baseline ED data on patient volume, were collected from the U.S. CDC;15 the 2007 edition of Westlaw, an online legal research tool for lawyers and legal professionals in the United States;16 and published literature,17 respectively.

Data Analysis

We calculated the percentage of sites providing each response to the items on the questionnaire. Missing data were not imputed, because only three respondents failed to complete the whole survey; their missing answers were simply treated as missing. Continuous variables obtained from sources other than the questionnaire (e.g., HIV prevalence, ED visit volume) were divided into quartiles. We then stratified all of these data by the presence or absence of rapid HIV testing at the site. All comparisons of binary variables used chi-square or Fisher’s exact tests. Mantel-Haenszel chi-square test for trend was used for ordinal variables (e.g., quartiles of HIV prevalence). We calculated odds ratios (ORs) and risk differences with 95% confidence intervals (CIs), as appropriate.

To assess temporal changes in use of HIV testing in academic EDs in the past decade, we compared responses between our questionnaire and a similar one administered in 1996.11 Both asked about typical physician actions regarding HIV testing in specific clinical scenarios. The scenarios for which we made comparisons were occupational exposure, suspected HIV, suspected STD, history of intravenous drug use, and general trauma from a motor vehicle collision. The comparisons are limited, however, because the 1996 questionnaire asked only if HIV testing occurred or did not occur. In contrast, our questionnaire gave six response options for each scenario (no testing or counseling, risk reduction counseling, verbal referral for HIV test, a written referral for testing, rapid testing, and standard testing). Respondents could choose all options that applied to what “would typically be done” in their EDs. For analysis and comparison purposes, we collapsed these response options into three groups: nontesting, written referral, and HIV test. Given the differences in the question format, and the fact that the actual data from the prior survey were not available at the individual-site level, formal tests of statistically significant changes over time were not conducted.

Results

We received 102 responses (80%) from the 128 academic EDs that were surveyed. Seventy-nine of the respondents (77%) had practiced EM for more than 10 years. The 2003 median visit volume of the surveyed EDs was 66,486 visits per year (interquartile range [IQR] = 34, 314–150,000). The 2004 median state-based rate of adults living with AIDS was 149.9 cases per 100,000 adults (IQR = 48.4–437.1 per 100,000). There was no pattern of nonresponse among hospitals by geography (dividing the country into Northeast, South, Midwest, West; p = 0.826) nor among hospitals situated in states with lower rates of adults living with AIDS (p = 0.076). However, the median number of annual visits at nonrespondents’ hospitals (47,310 visits) was smaller than the median number of annual visits of the respondents’ hospitals (66,485.5 visits) (p < 0.001).

Table 1 summarizes the survey results regarding current HIV testing policies at the study sites. Fifty-eight of the sites (57%, 95% CI = 47% to 66%) allow ordering of rapid HIV tests. Thirteen percent of the EDs already had standard guidelines stating that their providers should offer HIV screening to their patients as a routine preventive health measure, as suggested in the revised CDC recommendations. Some of the EDs had specific guidelines operationalizing when HIV testing should be performed. For example, 31% of the EDs specified that a rapid test should be used in the setting of occupational exposures. Fewer than 7% of EDs had policies specifying use of rapid testing in other clinical scenarios.

Table 1.   Current HIV Testing Practices at ED Sites (n = 102 of 128)
 Percentage95% CI
  1. HIV = human immunodeficiency virus; ED = emergency department; CI = confidence interval; AIDS = acquired immune deficiency syndrome; STD = sexually transmitted disease.

EDs where providers can order a rapid HIV test for any purpose5747%, 67%
EDs with policies/guidelines recommending
 Testing an employee postoccupational exposure3122%, 41%
 Testing a patient with an AIDS-defining illness requiring admission72%, 12%
 Testing a patient with an AIDS-defining illness not requiring admission51%, 9%
 Testing a patient with an uncomplicated STD40%, 8%
 Testing a patient with known intravenous drug use40%, 8%
 Testing a patient following a motor vehicle collision30%, 6%
 Screening of patients as a routine preventive health measure136%, 19%

Table 2 summarizes the prevalence of different policies and facility characteristics that might influence the presence of rapid HIV testing. Of the surveyed EDs, 82% offered at least one preventive health screening test and 66% had a private room that could be used for counseling. Few (9%) of the respondents’ EDs had HIV counselors available on site—a potential problem for future testing programs because 43% of these EDs are situated in states where pretest written consent is mandated by state statute (counseling can be conducted by any trained individual). Only 34 (59%) of the 58 EDs with rapid testing had the ability to link someone who tested positive with subspecialty HIV care.

Table 2.   Site Policies and Characteristics Potentially Influencing Availability of Rapid HIV testing (n = 102 of 128)
 Percentage95% CI
  1. HIV = human immunodeficiency virus; CI = confidence interval; ED = emergency department

  2. *The preventive health screenings that were inquired about were domestic violence, alcohol abuse, seatbelt safety, tobacco use, pneumonia vaccination, HIV, and “other, please specify.”

  3. †Calculated from total of agree/strongly agree response options from 5-point Likert scale.

  4. ‡States with this requirement include Hawaii, Massachusetts, Maine, Maryland, Michigan, Nebraska, New York, Pennsylvania, Rhode Island, and Wisconsin.16

Presence of at least one other preventive health screening test*8275%, 90%
Presence of external funding for ED program research, screening, treatment, etc.4636%, 56%
Presence of devoted space for private conversations with patients6656%, 75%
Presence of point person faculty member who promotes HIV testing in the ED1911%, 26%
Presence of on-site ED satellite lab3627%, 46%
Availability of HIV counselor in <1 hour†93%, 14%
Availability of interpreter in <1 hour†7061%, 79%
State statute for ED requires written informed consent prior to HIV testing in EDs‡3727%, 46%

We calculated the difference in prevalence of each facility-based characteristic between EDs with and without rapid HIV test availability. “Presence of other preventive health screening tests” was present in 88% of EDs with the capability to offer rapid testing in any situation and in 75% without this capability, a difference that, although moderate in size, did not reach statistical significance (p = 0.08). All other facility-based characteristics were not significantly different by rapid test availability (all p-values > 0.30). Similarly, facility and regional characteristics such as presence of state statutes requiring written informed consent (OR = 1.27, 95% CI = 0.57 to 2.85), ED visit volume (OR = 0.96, 95% CI = 0.67 to 1.35), and AIDS prevalence (OR = 0.79, 95% CI = 0.56 to 1.12) were not statistically different between sites with and without access to rapid HIV testing.

Table 3 summarizes respondents’ opinions about factors that most influence the presence of ED-based rapid HIV testing services. The most important factor was timely presence of an HIV counselor. The next four factors had overlapping CIs: ability to link patients who test positive with specialty care providers, presence of an HIV “champion” or point person among ED faculty, quick return of a rapid HIV test, and reasonable patient volume.

Table 3.   Facility Characteristics That Most Influence Presence of ED-based Rapid HIV Testing (n = 102 of 128)
 Percentage95% CI
  1. ED = emergency department; HIV = human immunodeficiency virus; CI = confidence interval.

  2. *”Reasonable” patient volume was intended to be subjectively defined by the respondent.

Timely presence of HIV counselor5040%, 60%
Ability to link patients to HIV specialty care providers3829%, 48%
Presence of an HIV “champion” or point person among ED faculty3223%, 42%
Quick return of rapid HIV test result3021%, 39%
Reasonable patient volume2618%, 35%
Access to an ED-based laboratory from which official results can be documented2314%, 31%
Presence of other preventive health screening tests169%, 23%
Potential for reimbursement for preventive health tests147%, 21%
Ability to have private conversations with patients147%, 21%
Access to external funding to support testing program125%, 18%
Reasonable patient:provider ratio*93%, 14%
Timely presence of interpreting services61%, 11%
High ratio of physician:nonphysician providers20%, 5%

In addition, respondents’ opinions were solicited regarding three other issues: 1) the effect of pretest counseling and consent requirements on patient flow, 2) the effect of waiting for a rapid test result on patient flow, and 3) the likelihood that their ED would offer rapid screening tests in the future. Each individual result was assessed to determine if it was affected by the presence or absence of ED-based rapid testing and, if so, if ED visit volume was also an important factor.

The presence of rapid testing only significantly affected the answers to the first opinion question in which 62% (95% CI = 52% to 71%) of all respondents felt that pretest consent and counseling would slow down patient flow. Specifically, for each 1-unit increase on a 5-point Likert scale measuring the strength of the belief that obtaining consent and providing counseling impedes patient flow, there was a 30% reduction in the odds of the site having rapid testing available (OR = 0.70; 95% CI = 0.51 to 0.97). The strength of this belief that obtaining consent and providing counseling for HIV testing impedes patient flow was not, however, associated with ED visit volume (OR = 0.79; 95% CI = 0.55 to 1.14).

The results from neither of the second two opinion questions appeared to be affected by the presence of rapid testing in the ED (p = 0.20 and p =0.89, respectively). As such, aggregated data for all respondents are reported: 35% (95% CI = 26% to 45%) of all respondents felt that waiting for the result of a rapid test would impede patient flow, and 33% (95% CI = 24% to 43%) of all respondents reported that their EDs were likely to offer routine nontargeted screening in the near future (i.e., would decide to follow CDC guidelines).

Figure 1 summarizes the HIV testing patterns elicited when respondents were provided with specific clinical scenarios. In the clinical scenario of an occupational exposure, 69% of respondents reported that providers in their ED typically tested the affected hospital employee for HIV. In the clinical scenario of an uncomplicated STD, 13% of respondents reported that providers in their ED typically performed an HIV test, and 26% typically provided a written referral to the patient. For the clinical scenario of a patient presenting with symptoms of an AIDS-defining disease, 36% of respondents reported that they would perform an HIV test.

Figure 1.

 ED providers’ HIV testing practices in scenarios of different suspicion of HIV. This histogram displays the percentage of sites whose providers reported different HIV testing and counseling practices in each of six specific clinical scenarios. Possible outcomes reported were no provider order for HIV counseling or testing, a written referral to outpatient testing, or HIV testing by rapid or standard means. Outcomes not reported include HIV risk reduction counseling and verbal referral to outpatient testing. Missing responses accounted for < 1% of the data for any individual question and were dropped from the calculation of the above statistics. Some of the categories do not total to 100% due to two reasons: 1) respondents had the option to choose more than one response for each question and 2) some of the outcomes were not presented for purposes of simplification. AIDS = acquired immune deficiency syndrome; STD = sexually transmitted disease; HIV = human immunodeficiency virus; ED = emergency department.

Discussion

Our national survey demonstrates that just over 50% of academic EDs have access to rapid HIV testing and that half of these EDs appear to restrict use of rapid HIV tests to situations involving occupational exposure. We found that few of the academic EDs we surveyed (13%; 95% CI = 6% to 19%) had a policy of routinely screening patients for HIV. According to responses to a clinical scenario, we also found that few respondents (4%; 95% CI = 0 to 8%) seem to have adopted routine screening as a practice (with CIs to the two questions overlapping). This limited amount of routine screening is not surprising given the ongoing controversy over the recent CDC recommendations to offer HIV testing routinely to all patients aged 13 to 64 years.6–9 Practical concerns also may be explanatory. Only 59% of the EDs who offered rapid tests in any situation had the ability to link a patient who tested positive to subspecialty care after discharge from the ED. This point is particularly noteworthy, since this factor ranked second among characteristics felt to influence the presence of a routine screening program.

Among this sample of U.S. EDs with academically accredited training programs, we did not observe an association between any of the hospital policies or characteristics that we measured and the presence of ED-based rapid HIV testing. Similarly, the regional prevalence of HIV was not associated with the presence of ED-based rapid HIV testing.

Our data can be compared to a nationally representative probability sample of 584 U.S. hospitals that found that only 52% of hospitals offered any rapid HIV tests in 2006 in any medical setting, and that only 21% more had plans to begin rapid testing in the immediate future.18 Similar to our data, the authors noted that half of the surveyed hospitals using the rapid tests did so for occupational exposures. However, in contrast to the greater than 50% of academic EDs that we found to have access to rapid HIV testing, that study determined that only 13% of all surveyed hospitals used rapid tests in their ED or urgent care centers at all. This discrepancy would be consistent with the idea that academic hospitals are earlier adopters of technology.

Another report has suggested that making a rapid test accessible in an ED could lead to more testing.19 However, our data suggest that wider use of the tests appears to be limited by ED-based guidelines that restrict testing to “in-house” use (e.g., occupational exposures). Such guidelines influence physicians’ decisions on when to use testing. For example, in a clinical scenario that should lead to high suspicion of a patient’s engagement in unprotected sex (a suspected STD), only 4% of EDs have guidelines that support testing, only 13% of sites offered HIV testing routinely, and 26% provided a written referral for outpatient testing (a notable approach, given that written referrals for testing have proven generally ineffective).20 More than 50% of the respondents surveyed did not consistently offer HIV testing in any form to such a patient.

These numbers have changed only minimally since 1996, when a similar survey was conducted.11 That study reported that 54% of EDs tested for HIV in situations of occupational exposure, 36% tested patients in whom they had a high suspicion of HIV, 3% tested patients with uncomplicated STDs (although 51% reported that the majority of their colleagues would provide a written referral), and approximately 10% tested patients with a history of intravenous drug use. The overall low level of HIV testing among high risk patients is consistent with the 10% rate of testing of patients with STDs found in a 2002 survey of 154 ED providers within one county in the Northeastern United States.21

Reported beliefs about factors thought most important to promoting rapid HIV testing were revealing. All respondents agreed on the importance of having HIV counselors perform pre- and posttest counseling. They also identified the ability to connect patients who test positive to specialty care providers as an important factor underlying the availability of rapid HIV testing. As noted above, though, only 59% of the EDs with rapid testing reported that they had the ability to make this connection, suggesting that this unmet need may deter some providers from screening in routine situations.

Thirty-three percent of sites indicated that they would be likely to follow CDC recommendations for routine nontargeted HIV screening within the next 2 to 3 years. Expanding routine screening in this manner is critical, since traditional risk factor-based testing is not sufficient to locate undetected carriers of the disease.10,22 High numbers of infected persons are less than 20 years old, members of racial or ethnic minority groups, heterosexual, less educated, and urban dwelling.23,24 Because these individuals are often uninsured, they lack access to regular health care and may also be unaware of their risk of acquiring and transmitting HIV infection.25 Effective efforts to halt the spread of the epidemic should focus on outreach to such hard-to-reach individuals.

Emergency departments, where many of the uninsured turn for their primary health needs, have great potential to serve as a site for outreach activities. One recently published study found that an ED screening program successfully detected HIV infection in patients who might not have been tested through risk- or symptom-based strategies and also was able to link these patients with further care.26 Although the report notes that diagnostic yield was higher among provider-referred patients, the use of routine screening led to detection of carriers earlier in the course of disease. Another study of three major urban centers offered similar success.27 It piloted different means of offering rapid HIV tests—two centers used counselors, and a third used existing staff. The centers that used counselors offered routine screening tests to fewer people overall (<4% of all patients), but more of those people accepted testing (90%). The center that used existing staff offered routine screening tests to more than half of their patients, of whom 39% were actually tested. Most important, 85 (88%) of the 97 people newly diagnosed with HIV infection over the 14 months of study were successfully linked to appropriate health services.

Limitations

While our response rate of 80% is high among physician surveys, nonresponse bias remains a concern. Information bias may also be a concern, because we made the assumption that each respondent was very familiar with the testing practices in their ED. We attempted to limit the effect of the individual’s perceptions by focusing on concrete facts of the EDs’ stated policies, general practices, and facilities, as well as augmenting the subjective responses with objective data derived from external sources. Even so, some bias may have been introduced into the results to the degree that individuals misreported the status of testing or testing-related policies. The generalizability of our results is limited by the study’s use of participants in a particular research network and not of locally identified clinical leaders. However, the academic centers with residency programs where the respondents work may act as the opinion leaders in their regions. By training residents, each is in a position to establish practice patterns for future generations of EM physicians and therefore may influence practice patterns more broadly. Accordingly, we believe that academic EDs represent an important subset of U.S. EDs that merit study.

Conclusions

The majority of academic EDs in the United States have not yet implemented CDC guidelines on HIV testing. Barriers to use of the tests include lack of adoption of rapid testing, department policies restricting testing to occupational exposures, and lack of uptake of testing among ED physicians (even in situations with high pre-test probability like STD exposure or intravenous drug use). Concerns that screening programs are not feasible in EDs with high patient volume were significantly less common in EDs that already had rapid HIV testing programs in place. However, access to HIV counselors, and having the capacity to link newly identified HIV-positive patients to specialty HIV care, remained important factors that limited availability of any rapid HIV testing, even in these sites (as well as in sites without rapid HIV testing at all). Along with quick return of the test result, both of these characteristics appear to be crucial elements in determining whether ED-based rapid HIV testing or screening programs are implemented. Further investigations and pilot programs are needed to more precisely define and address the conceptual and practical gaps that lead to the divide between the CDC recommendations for ED-based screening and the practice realities described in this study.

The authors acknowledge Sharon Wilson, MD, of UC Davis Medical Center; Yu-Hsiang Hsieh, MD, of Johns Hopkins Medical Center; and Kate E. Delaney, BA, Ashley Sullivan, MS, MPH, and Sunday Clark, MPH, ScD, all of Massachusetts General Hospital. Dr. Wilson and colleagues performed the 1996 study referenced throughout the paper and agreed to share their original survey instrument with us. Dr. Hsieh, Ms. Delaney, Ms. Sullivan, and Dr. Clark all provided technical support instrumental to the completion of the study.

Ancillary